Theoretical Analysis of Factors Controlling Pd-Catalyzed

Dec 28, 2009 - J. Am. Chem. Soc. All Publications/ ...... Nelly Rameau , Baptiste Russo , Stéphane Mangematin , Catherine Pinel , Laurent Djakovitch...
1 downloads 0 Views 3MB Size
Download PDF
Published on Web 12/28/2009

Theoretical Analysis of Factors Controlling Pd-Catalyzed Decarboxylative Coupling of Carboxylic Acids with Olefins Song-Lin Zhang,† Yao Fu,*,† Rui Shang,† Qing-Xiang Guo,† and Lei Liu*,‡ Departments of Chemistry, UniVersity of Science and Technology of China, Hefei 230026, China, and Tsinghua UniVersity, Beijing 100084, China Received September 3, 2009; E-mail: [email protected]; [email protected]

Abstract: Transition-metal-catalyzed decarboxylative coupling presents a new and important direction in synthetic chemistry. Mechanistic studies on decarboxylative coupling not only improve the understanding of the newly discovered transformations, but also may have valuable implications for the development of more effective catalyst systems. In this work, a comprehensive theoretical study was conducted on the mechanism of Myers’ Pd-catalyzed decarboxylative Heck reaction. The catalytic cycle was found to comprise four steps: decarboxylation, olefin insertion, β-hydride elimination, and catalyst regeneration. Decarboxylation was the rate-limiting step, and it proceeded through a dissociative pathway in which Pd(II) mediated the extrusion of CO2 from an aromatic carboxylic acid to form a Pd(II)-aryl intermediate. Further analysis was conducted on the factors that might control the efficiency of Myers’ decarboxylative Heck reaction. These factors included Pd salts, ligands, acid substrates, and metals. (1) Regarding Pd salts, PdCl2 and PdBr2 were worse catalysts than Pd(TFA)2, because the exchange of Cl or Br by a carboxylate from Pd was thermodynamically unfavorable. (2) Regarding ligands, DMSO provided the best compromise between carboxyl exchange and decarboxylation. Phosphines and N-heterocarbenes disfavored decarboxylation because of their electron richness, whereas pyridine ligands disfavored carboxyl exchange. (3) Regarding acid substrates, a good correlation was observed between the energy barrier of R-COOH decarboxylation and the R-H acidity. Substituted benzoic acids showed deviation from the correlation because of the involvement of π(substituent)-σ(Cipso-Pd) interaction. (4) Regarding metals, Ni and Pt were worse catalysts than Pd because of the less favorable carboxyl exchange and/or DMSO removal steps in Ni and Pt catalysis.

1. Introduction

Transition-metal-catalyzed decarboxylative cross-coupling using carboxylic acids as aryl sources has received considerable attention recently.1 Compared to conventional catalytic couplings2 where expensive unstable organometallic reagents are often required, decarboxylative coupling utilizes readily available and stable carboxylic acids as substrates, thus precluding the need for preparation of sensitive organometallic reagents. An additional advantage is that decarboxylative coupling generates CO2 without producing toxic metal halides. Pioneering work by Nilsson showed that stoichiometric Cu promoted decarboxylative coupling of aromatic acids with ArI.3 Cohen et al. found that with phenanthroline ligand Cu(I) could mediate decarboxylation of aromatic acids, producing aryl-Cu that could be protonated to yield arenes.4 Recently, Goossen et al. discovered elegant protocols for the decarboxylative coupling †

University of Science and Technology of China. Tsinghua University. (1) (a) Baudoin, O. Angew. Chem., Int. Ed. 2007, 46, 1373. (b) Goossen, L. J.; Rodriguez, N.; Goossen, K. Angew. Chem., Int. Ed. 2008, 47, 3100. (2) Metal-Catalyzed Cross-Coupling Reactions; de Meijere, A., Diederich, F., Eds.; Wiley: Weinheim, Germany, 2004. (3) (a) Nilsson, M. Acta Chem. Scand. 1966, 20, 423. (b) Bjo¨rklund, C.; Nilsson, M. Acta Chem. Scand. 1968, 22, 2585. (4) (a) Cohen, T.; Schambach, R. A. J. Am. Chem. Soc. 1970, 92, 3189. (b) Cohen, T.; Berninger, R. W.; Wood, J. T. J. Org. Chem. 1978, 43, 837. ‡

638

9

J. AM. CHEM. SOC. 2010, 132, 638–646

of benzoic acids with ArBr via a Cu/Pd bimetallic catalysis.5 Cu was used for decarboxylation to produce aryl-Cu. It transmetalates with aryl-Pd(II) generated from oxidative addition of ArBr to Pd(0), ending up with a bis-arylated Pd(II) species. Finally, reductive elimination gives the biaryl product. Goossen’s reaction has been nicely extended to ArCl6 and ArOTf7 as the electrophiles. Moreover, R-oxo carboxylates could be transformed under the bimetallic catalysis to give aryl ketones through decarboxylative coupling.8 In addition to Cu, Pd by itself can also promote decarboxylative coupling. In 2002, Myers et al. discovered an interesting Heck-type reaction between aromatic acids and olefins catalyzed by Pd(TFA)2 (TFA ) CF3CO2-) (eq 1).9 Kinetic analysis indicated that decarboxylation was the rate-limiting step of the (5) (a) Goossen, L. J.; Deng, G.; Levy, L. M. Science 2006, 313, 662. (b) Goossen, L. J.; Rodriguez, N.; Melzer, B.; Linder, C.; Deng, G.; Levy, L. M. J. Am. Chem. Soc. 2007, 129, 4824. (c) Goossen, L. J.; Melzer, B. J. Org. Chem. 2007, 72, 7473. (d) Goossen, L. J.; Knauber, T. J. Org. Chem. 2008, 73, 8631. (6) Goossen, L. J.; Zimmermann, B.; Knauber, T. Angew. Chem., Int. Ed. 2008, 47, 7103. (7) Goossen, L. J.; Rodriguez, N.; Linder, C. J. Am. Chem. Soc. 2008, 130, 15248. (8) Goossen, L. J.; Rudolphi, F.; Oppel, C.; Rodriguez, N. Angew. Chem., Int. Ed. 2008, 47, 3043. (9) (a) Myers, A. G.; Tanaka, D.; Mannion, M. R. J. Am. Chem. Soc. 2002, 124, 11250. (b) Tanaka, D.; Myers, A. G. Org. Lett. 2004, 6, 433. (c) For a recent improvement of the Myers reaction, see: Hu, P.; Kan, J.; Su, W.; Hong, M. Org. Lett. 2009, 11, 2341. 10.1021/ja907448t  2010 American Chemical Society

Factors Controlling Decarboxylative Coupling Scheme 1. Proposed Mechanism for Myers’ Decarboxylative Cross-Coupling

catalytic cycle.10 The proposed aryl-Pd(II) intermediate from decarboxylation was captured and characterized by NMR and X-ray methods. This aryl-Pd(II) intermediate could undergo a facile, stoichiometric reaction with an olefin to yield the coupling product. On the basis of the results, a mechanism (Scheme 1) was proposed in which initial rate-limiting decarboxylation was followed by classical mechanistic steps of Heck reactions, i.e., olefin insertion, β-hydride elimination, and catalyst regeneration.

ARTICLES

boxylation of benzoic acids.18 On the other hand, almost no theoretical study has been conducted to understand the mechanistic details of Pd-catalyzed decarboxylative coupling. Only Myers’ study provided some spectroscopic and crystallographic evidence for some intermediates in the proposed process.10 Even so, the structural and energetic details about how these intermediates transform to each other remain largely unknown. Here we report a thorough theoretical study on Pd-catalyzed decarboxylative couplings. Because Myers’ decarboxylative Heck-type coupling stands for the best experimentally characterized system to date, we choose to focus on this particular reaction so that the calculation results can be compared to the experiment. Through the study we have obtained detailed structural and energetic information about each step of the decarboxylative coupling. More importantly, through systematic analysis of the catalysts, substrates, and ligands, the study sheds important insights into the various factors controlling the activation barrier of the decarboxylation process. These results may have valuable implications for the development of new, more effective catalyst systems for decarboxylative couplings. 2. Methods

After Myers’ pioneering work, Forgione et al. reported Pdcatalyzed decarboxylative coupling of heteroarenecarboxylic acids with ArBr.11 Lee et al. reported Pd-catalyzed decarboxylative coupling of alkynecarboxylic acids with ArBr.12 Becht et al. reported Pd-catalyzed decarboxylative coupling of benzoic acids with Ar2I+OTf-.13 Wu14 and Crabtree15 groups reported Pd-catalyzed decarboxylative coupling of cinnamyl acids and benzoic acids with ArI. In related studies, Tunge et al. developed novel methods for the substitution of allylic terminus via Pdcatalyzed intramolecular decarboxylative reaction of allyl carboxylic esters.16 Our own work in this area was the discovery of novel Pd(II)-catalyzed synthesis of aromatic esters through decarboxylative coupling of oxalate monoesters with aryl halides.17 Despite the remarkable advances and great contemporary interest in catalytic decarboxylative couplings, many mechanistic details of the decarboxylation process remain ambiguous. As to Cu-based systems, Goossen et al. recently performed DFT calculations to analyze Cu(I)(phenanthroline)-mediated decar(10) Tanaka, D.; Romeril, S. P.; Myers, A. G. J. Am. Chem. Soc. 2005, 127, 10323. (11) Forgione, P.; Brochu, M.-C.; St-Onge, M.; Thesen, K. H.; Bailey, M. D.; Bilodeau, F. J. Am. Chem. Soc. 2006, 128, 11350. (12) (a) Moon, J.; Jeong, M.; Nam, H.; Ju, J.; Moon, J. H.; Jung, H. M.; Lee, S. Org. Lett. 2008, 10, 945. (b) Moon, J.; Jang, M.; Lee, S. J. Org. Chem. 2009, 74, 1403. (13) (a) Becht, J.-M.; Le Drian, C. Org. Lett. 2008, 10, 3161. (b) Becht, J.-M.; Catala, C.; Le Drian, C.; Wagner, A. Org. Lett. 2007, 9, 1781. (14) (a) Wang, Z. Y.; Ding, Q. P.; He, X. D.; Wu, J. Org. Biomol. Chem. 2009, 7, 863. (b) Wang, Z. Y.; Ding, Q. P.; He, X. D.; Wu, J. Tetrahedron 2009, 65, 4635. (15) Voutchkova, A.; Coplin, A.; Leadbeater, N. E.; Crabtree, R. H. Chem. Commun. 2008, 6312. (16) (a) Rayabarapu, D. K.; Tunge, J. A. J. Am. Chem. Soc. 2005, 127, 13510. (b) Burger, E. C.; Tunge, J. A. J. Am. Chem. Soc. 2006, 128, 10002. (c) Waetzig, S. R.; Rayabarupu, D. K.; Weaver, J. D.; Tunge, J. A. Angew. Chem., Int. Ed. 2006, 45, 4977. (d) Waetzig, S. R.; Tunge, J. A. J. Am. Chem. Soc. 2007, 129, 14860. (17) Shang, R.; Fu, Y.; Li, J. B.; Zhang, S. L.; Guo, Q. X.; Liu, L. J. Am. Chem. Soc. 2009, 131, 5738.

All calculations were performed with Gaussian 03.19 B3LYP method was used.20 Geometry optimization was conducted with the 631LAN basis set (i.e., Lanl2dz for Ni, Pd, or Pt, and 6-31G(d) for the other elements).21 Frequency analysis was conducted at the same level of theory to verify the stationary points to be real minima or saddle points and to get the thermodynamic energy corrections. For each saddle point, the intrinsic reaction coordinate (IRC) analysis22 was carried out to confirm that it connected the correct reactant and product on the potential energy surface. Natural population analysis (NPA) was performed also at the same level of theory.23 Single-point energy calculations were performed on the stationary points by using a larger basis set, i.e., SDD for Ni, Pd, or Pt and 6-311+G(d, p) for the other elements. Solvent effect (solvent ) DMF) was calculated by using self-consistent reaction field method24 with CPCM solvation model25 and UAHF radii. Single-point energies corrected by Gibbs free energy corrections and solvation energies were used to describe the reaction energetics throughout the study. (18) Goossen, L. J.; Thiel, W. R.; Rodriguez, N.; Linder, C.; Melzer, B. AdV. Synth. Catal. 2007, 349, 2241. (19) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; et al. Gaussian 03, reVision D.01; Gaussian, Inc.: Pittsburgh, PA, 2004. (20) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (21) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 299. (22) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523. (23) (a) Reed, A E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985, 83, 735. (b) Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F. NBO Version 3.1 in the Gaussian 98 Package; University of Wisconsin: Madison, WI, 1990. (24) Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. J. Comput. Chem. 2003, 24, 669. (25) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. ReV. 2005, 105, 2999. (26) For recent reviews on Heck reactions, see: (a) Beletskaya, I. P.; Cheprakov, A. V. Chem. ReV. 2000, 100, 3009. (b) Dounay, A. B.; Overman, L. E. Chem. ReV. 2003, 103, 2945. (27) Some previous theoretical studies on Heck couplings and related Pdcatalyzed coupling reactions: (a) Deeth, R. J.; Smith, A.; Hii, K. K.; Brown, J. M. Tetrahedron Lett. 1998, 39, 3229. (b) Lin, B. L.; Liu, L.; Fu, Y.; Luo, S. W.; Chen, Q.; Guo, Q. X. Organometallics 2004, 23, 2114. (c) Deeth, R. J.; Smith, A.; Brown, J. M. J. Am. Chem. Soc. 2004, 126, 7144. (d) Kozuch, S.; Shaik, S. J. Am. Chem. Soc. 2006, 128, 3355. (e) Goossen, L. J.; Koley, D.; Hermmann, H. L.; Thiel, W. Organometallics 2005, 24, 2398. (f) Goossen, L. J.; Koley, D.; Hermann, H. L.; Thiel, W. J. Am. Chem. Soc. 2005, 127, 11102. (g) Goossen, L. J.; Koley, D.; Hermann, H. L.; Thiel, W. Organometallics 2006, 25, 54. (h) Surawatanawong, P.; Fan, Y.; Hall, M. B. J. Organomet. Chem. 2008, 693, 1552. (i) Surawatanawong, P.; Hall, M. B. Organometallics 2008, 27, 6222. J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

639

ARTICLES

Zhang et al.

Scheme 2. Detailed Reaction Pathway for Myers’ Decarboxylative Heck Reaction Supported by Current Theoretical Analysis (Dissociative

Decarboxylation Pathway)

3. Catalytic Cycle of the Myers Reaction

The catalytic cycle of the model reaction between benzoic acid and styrene (eq 2) is proposed on the basis of Myers’ study and the established mechanism for Pd-catalyzed Heck reaction (Scheme 2).26,27 There are four steps in the cycle: decarboxylation (CP0 f CP3), olefin insertion (CP4 f CP5), β-hydride elimination (CP6 f CP8), and catalyst regeneration (CP8 f CP0).

3.1. Decarboxylation. Decarboxylation starts with trans(DMSO)2Pd(TFA)2 (CP0). This complex has three possible isomers depending on how the DMSO molecules are bound to Pd. Our calculations (in DMF) indicate that the bis-S-bound isomer (CP0-ss) is the least stable (+8.4 kcal/mol), whereas CP0-os (mono-O-mono-S-bound, +0.1 kcal/mol) and CP0-oo (bis-O-bound, +0.0 kcal/mol) are nearly isoenergetic. Thus, dynamic exchange between S- and O-bound forms28-30 must be considered at every stage of the decarboxylative coupling process. Furthermore, trans-(DMSO)2Pd(TFA)2 is found to be much more stable than its cis isomer by 3.4 kcal/mol. Note that the crystal structure of (DMSO)2Pd(TFA)2 also corresponds to (28) Some other examples for the use of DMSO as possible coordinating ligand in Pd-catalyzed processes include the following: (a) Larock, R. C.; Hightower, T. R. J. Org. Chem. 1993, 58, 5298. (b) Chen, M. S.; White, M. C. J. Am. Chem. Soc. 2004, 126, 1346. (c) Zhou, C.; Larock, R. C. J. Am. Chem. Soc. 2004, 126, 2302. (29) (a) Steinhoff, B. A.; Fix, S. R.; Stahl, S. S. J. Am. Chem. Soc. 2002, 124, 766. (b) Steinhoff, B. A.; Stahl, S. S. J. Am. Chem. Soc. 2006, 128, 4348. (30) Zierkiewicz, W.; Privalov, T. Organometallics 2005, 24, 6019. 640

J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

the trans isomer but it exists as the CP0-os isomer.31 The slight difference between the crystal and computational structures may be attributed to the crystal packing force. CP0 undergoes carboxyl exchange with benzoic acid, producing intermediate CP1. The free energy change of this step is calculated to be +3.7 kcal/mol. Note that for CP1 the bisO-bound isomer (i.e., CP1-oo) is again the most stable (Figure 1). From CP1-oo (note that in the following discussion we will only mention the most stable isomer), either the associative or dissociative pathway can take place for the extrusion of CO2. Both possibilities are examined as follows (Scheme 3). Associative Pathway. In this pathway the two DMSO ligands remain attached to Pd during decarboxylation. Transition state TSassoc-oo is successfully located in which the two DMSOs are both O-bound (Figure 1). The energy barrier for this step is +41.2 kcal/mol as calculated from CP0-oo. In TSassoc-oo, the Cipso atom of the phenyl ring, an O atom of the leaving CO2, the O atom of one DMSO, and an O atom of TFA form the coordination plane (Figure 1). Our calculation indicates that it is energetically more favorable for DMSO instead of TFA to occupy the trans position of Cipso. The immediate product after CO2 removal is a four-coordinated Pd complex (PDcis-os) with two DMSOs cis to each other. PDcis-os is expected to transform to its more stable trans-isomer PDtrans-os. Note that Myers et al. obtained the crystal structure of the trans-aryl-Pd(II) trifluoroacetate intermediate carrying two DMSOs, which was also found to be a trans complex.10 However, in Myers’ crystal the two DMSOs were both S-bound, whereas our calculations (31) Bancroft, D. P.; Cotton, F. A.; Verbruggen, M. Acta Crystallogr., Sect. C 1989, 45, 1289.

Factors Controlling Decarboxylative Coupling

ARTICLES

Figure 1. Key intermediates and transition states in associative and dissociative decarboxylation pathway. Scheme 3. Associative and Dissociative Pathway for Decarboxylationa

a

Values in parentheses are relative Gibbs free energies in DMF (kcal/mol).

indicate that PDtrans-os is more stable than PDtrans-ss by 1.8 kcal/mol in DMF. Dissociative Pathway. In this pathway only one DMSO remains attached to Pd during decarboxylation. To do so, CP1oo releases one DMSO, whose left position is immediately occupied by the phenyl ring of PhCO2H through a η2 binding mode. This process produces an active intermediate CP2-o and its isomer CP2′-o (Scheme 3). Both CP2-o and CP2′-o can extrude CO2 through transition state TS1-o or TS1′-o. The structural difference between TS1-o and TS1′-o is whether the Cipso atom is trans or cis to DMSO (Figure 1). Our calculation shows that TS1′-o is less stable than TS1-o by 1.8 kcal/mol in DMF. Therefore, dissociative decarboxylation should proceed through TS1-o. It is interesting to note that, in TS1-o, four atoms (i.e., Pd, Cipso, C2, and O2) stay in the same plane that is perpendicular to the phenyl ring. This pathway has a relatively lower activation barrier of +29.0 kcal/mol. The immediate product of the dissociative decarboxylation is CP3-o, which can

exchange CO2 with DMSO or PhCHdCH2 to produce PDtransos or CP4-o respectively. 3.2. Olefin Insertion. Olefin insertion begins with CP4-o (Figure 2). CP4-o inserts the coordinated olefin into the aryl-Pd bond to produce CP5-o through a four-membered cyclic transition state TS2-o. The energy barrier for this step is +15.6 kcal/mol. In CP4-o the olefin is perpendicular to the Pd coordination plane, whereas in TS2-o the olefin lies in the Pd coordination plane. After olefin insertion, CP5-o isomerizes to CP5-s and becomes more stable by 1.3 kcal/mol. In CP5-s, the Cipso atom remains coordinated to the Pd center. This observation is in agreement with Myers’ crystal structure for a norbornene adduct of aryl-Pd(II)-TFA, which also turns out to be a S-bound complex with DMSO occupying the trans position of Cipso.10 The Pd-Cipso and Pd-CR distances in Myers’ crystal are 2.24 and 2.03 Å, as compared to 2.40 and 2.09 Å in CP5-s. 3.3. β-Hydride Elimination. To undergo β-hydride elimination, CP5-s rotates its phenyl group around the CR-Cβ bond J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

641

ARTICLES

Zhang et al.

Figure 2. Key intermediates and transition state in olefin insertion.

Figure 3. Key intermediates and transition state in β-hydride elimination.

so that one of the benzylic H atoms lies in the Pd-CR-Cβ plane. This gives intermediate CP6-o, featuring an agostic interaction of β-C-H to Pd center (Figure 3). From CP6-o, β-hydride elimination occurs to produce CP7-o through transition state TS3-o. The energy barrier for this step is only +5.4 kcal/mol as calculated from CP5-s. In both TS3-o and CP7-o, the hydride H occupies the trans position of DMSO. Furthermore, in CP7-o the olefin bond of stilbene is perpendicular to the coordination plane of Pd. 3.4. Catalyst Regeneration. In the last step, displacement of stilbene by DMSO releases the coupling product and generates a bis(DMSO)-Pd hydride CP8-oo. Then, the active catalyst (i.e., trans-CP0-oo) is regenerated from CP8-oo. This process has been recently studied by both experimental and computational methods.32 Two pathways have been proposed. One is the direct insertion pathway in which the oxidant directly inserts into the Pd-H bond to produce (X)Pd(II)-hydroperoxide. This intermediate is protonated by HX to give PdX2 and hydroperoxide. The other one is the Pd(0) pathway, in which an initial reductive elimination of HX from Pd(H)(X) occurs to produce Pd(0) that is oxidized by the oxidant to become PdX2. It has been shown that, for labile, monodentate ligands such as pyridine, the Pd(0) pathway is favored. Because in Myers’ reaction the ligand is DMSO, we consider the Pd(0) pathway to be the favored one.33 (32) For recent reviews on Pd oxidations, see: (a) Stahl, S. S. Angew. Chem., Int. Ed. 2004, 43, 3400. (b) Gligorich, K. M.; Sigman, M. S. Angew. Chem., Int. Ed. 2006, 45, 6612. (c) Piera, J.; Backvall, J.-E. Angew. Chem., Int. Ed. 2008, 47, 3506. (33) For some recent studies on this subject, see: (a) Landis, C. R.; Morales, C. M.; Stahl, S. S. J. Am. Chem. Soc. 2004, 126, 16302. (b) Popp, B. V.; Stahl, S. S. J. Am. Chem. Soc. 2007, 129, 4410. (c) Konnick, M. M.; Stahl, S. S. J. Am. Chem. Soc. 2008, 130, 5753. (d) Keith, J. M.; Goddard, W. A., III. J. Am. Chem. Soc. 2009, 131, 1416. (e) Popp, B. V.; Stahl, S. S. Chem.sEur. J. 2009, 15, 2915. 642

J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

Our calculation shows that reductive elimination of CF3COOH from CP8-oo has an activation barrier of +8.7 kcal/ mol and is exothermic by -12.0 kcal/mol. Then, oxidation of Pd(0)(DMSO)2 by Ag2CO3 in the presence of trifluoroacetic acid gives (DMSO)2Pd(II)(TFA)2. This oxidation is slightly endothermic by +1.5 kcal/mol. Overall, catalyst regeneration from CP8-oo to trans-CP0-oo is exothermic by -10.4 kcal/mol and, therefore, thermodynamically favorable. 3.5. Overall Catalytic Cycle and Rate-Limiting Step. Figure 4 shows the free energy profile for the catalytic cycle of Myers’ decarboxylative Heck reaction. The part marked in blue describes the associative pathway, while the part in red describes the dissociative pathway. According to the profile, decarboxylation proceeds through the dissociative pathway with an activation barrier of +29.0 kcal/mol. Compared to decarboxylation, olefin insertion, β-hydride elimination, and catalyst regeneration are all facile steps. Thus, decarboxylation is the rate-determining step for Pd-catalyzed decarboxylative coupling of arenecarboxylic acids with alkenes. 4. Factors Controlling the Efficiency of Decarboxylative Coupling

In the catalytic cycle we identify that decarboxylation is the rate-determining step. In this part, the various factors controlling the decarboxylation step are discussed, which include the anionic ligand, the neutral ligand, substrate, and metal. As shown in Figure 4, the overall activation barrier for decarboxylation can be affected by the energy costs of three steps: (1) the energy required for carboxyl exchange (CP0 f CP1), (2) the energy required for dissociation of one DMSO from Pd (CP1 f CP2), and (3) the activation energy required for the extrusion of CO2 (CP2 f TS1). To better understand why the decarboxylation efficiency changes

Factors Controlling Decarboxylative Coupling

ARTICLES

Figure 4. Free energy profile for the catalytic cycle of Myers’ decarboxylative Heck reaction in DMF.

Figure 5. Fragment orbital analysis for TS1.

in different catalytic systems, the effect of each factor on the three steps is analyzed. 4.1. Orbital Analysis of the Decarboxylation Transition State. Before studying each factor, we analyze the orbital

interactions for the extrusion of CO2 from the Pd complex. Thus, we divide TS1 into two fragments (Figure 5), i.e., Pd(DMSO)(TFA)(Ph) and the deformed CO2 moiety. Calculation shows that the LUMO of the Pd(DMSO)(TFA)(Ph) fragment is mainly composed of the Pd 4dx2-y2 orbital and some contribution from the σ*(Pd-Cipso) antibond. The Pd 4dx2-y2 orbital is highly asymmetric presumably due to the stronger trans influence of the anionic TFA group than the neutral DMSO. The HOMO of the deformed CO2 is a π bond. It overlaps with the LUMO of Pd(DMSO)(TFA)(Ph) in the manner shown in Figure 5.

From the orbital interactions we can draw the following conclusions. First, an electron-deficient Pd center leads to a lowlying LUMO orbital that can interact more strongly with the π bond of deformed CO2. Thus, electron-deficient ligands, rather than electron-rich ones, are expected to facilitate decarboxylation. Second, anionic ligands with a stronger trans influence can stabilize the decarboxylation transition state because a stronger desymmetrization of the Pd 4dx2-y2 orbital can bring about a more favorable HOMO/LUMO interaction. Third, from the precomplex CP2 to TS1, the NPA charge on the Cipso atom changes from -0.168 to -0.319 while the NPA charge on the carboxyl carbon increases from +0.814 to +0.934. Thus, any factor that stabilizes the accumulating negative charges on Cipso or the positive charge on the carboxyl carbon can contribute to the stabilization of TS1, J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

643

ARTICLES

Zhang et al.

Table 1. Effect of Anionic Ligands on the Decarboxylation Step (kcal/mol)

a

X

∆G(CP0 f CP1)

∆G(CP1 f CP2)

∆Gq(CP2 f TS1)

∆Gqdecarboxylationa

TFA Cl Br

3.7 20.4 23.2

7.8 4.3 4.2

17.5 15.7 15.4

+29.0 +40.4 +42.8

∆Gqdecarboxylation ) ∆G(CP0 f CP1) + ∆G(CP1 f CP2) + ∆Gq(CP2 f TS1).

facilitating the decarboxylation. These conclusions can be used to explain the calculation results as discussed below. 4.2. Effect of Anionic Ligands on Decarboxylation. Myers and co-workers found that trifluoroacetate played a key role in the decarboxylative palladation reaction, because other Pd salts such as PdCl2 and PdBr2 were ineffective or gave inferior results.10 Moreover, added bromide completely inhibited decarboxylation.10 To understand these observations, PdCl2 and PdBr2 are examined to evaluate the effect of anionic ligands on decarboxylation (Table 1). According to the orbital analysis, the anionic ligand with a stronger trans influence can stabilize the transition state of decarboxylation (namely, TS1). Indeed, the ∆Gq(CP2 f TS1) values for Cl and Br are lower than that for TFA by ca. 2-3 kcal/ mol, because Cl and Br exert a stronger trans influence than TFA. However, it is found that, in the carboxyl exchange step, the ∆G(CP0 f CP1) value is only +3.7 kcal/mol for TFA, as compared to +20.4 and +23.2 kcal/mol for Cl and Br. Consequently, the overall activation barrier of decarboxylation is very high (over +40 kcal/mol) for PdCl2 and PdBr2. This explains why PdCl2 and PdBr2 provided inferior results in decarboxylation. 4.3. Effect of Neutral Ligand on Decarboxylation. Several common types of neutral ligands including phosphine, phosphite, pyridine, and N-heterocarbene are examined for their effects on the decarboxylation step (Table 2). It is found, not without surprise, that DMSO gives the lowest activation barrier as compared to all the above popular ligands. This calculation result is in agreement with the experimental observation that DMSO is thus far the only appropriate ligand system for the decarboxylative Heck reaction.9,10 A phosphine or phosphite ligand (e.g., PMe3 or P(OMe)3) is found to be detrimental to decarboxylation as reflected by the high activation barriers. The reason is 2-fold: (1) Phosphine or phosphite ligands are relatively electron-rich and, therefore, generate a less electron-deficient Pd(II) center. This would increase the energy cost for the extrusion of CO2 in the transition from CP2 toTS1 (27.5 kcal/mol for PMe3 and 26.1 kcal/mol for P(OMe)3, vs 17.5 kcal/ mol for DMSO). (2) Phosphine and phosphite ligands coordinate to Pd more strongly, so that the ligand dissociation step (i.e., CP1f CP2) also becomes highly unfavorable (13.9 kcal/mol for PMe3 and 10.4 kcal/mol for P(OMe)3, vs 7.8 kcal/mol for DMSO). Compared to phosphine and phosphite, pyridine is less electronrich, and it does not coordinate to Pd very strongly. In agreement with this argument, we find that the energy costs of the ligand dissociation (CP1 f CP2) and CO2 extrusion (CP2 f TS1) steps are relatively low for pyridine (8.9 and 19.0 kcal/mol, vs 7.8 and 17.5 kcal/mol for DMSO). However, it is found that the energy cost of the carboxyl exchange step (CP0 f CP1) is fairly high 644

J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

Table 2. Effect of Neutral Ligand on Decarboxylation of PhCOOH (kcal/mol)a

a ∆Gqdecarboxylation ) ∆G(CP0 f CP1) + ∆G(CP1 f CP2) + ∆Gq(CP2 f TS1).

for pyridine (10.2 kcal/mol vs 3.7 kcal/mol for DMSO). As a result, pyridine has a much higher activation barrier of decarboxylation than DMSO. Finally, an N-heterocarbene ligand is found to be most unfavorable for Pd-mediated decarboxylation. Its electron richness and its strong coordination ability to Pd lead to very high energy costs for the ligand dissociation (CP1 f CP2, 24.5 kcal/mol) and CO2 extrusion (CP2 f TS1, 26.7 kcal/mol) steps. Besides, it also has a high energy cost for the carboxyl exchange step (CP0 f CP1, 11.0 kcal/mol). 4.4. Effect of Carboxylic Acid Substrates on Decarboxylation.

A number of carboxylic acids are examined to study the effect of substrates on decarboxylation. According to the results in Table 3, the following predictions can be made. First, alkynecarboxylic acids can undergo facile decarboxylation as indicated by the low activation barrier of propynecarboxylic acid (+16.3 kcal/mol). This prediction is supported by the study of Lee,12 who showed that alkynecarboxylic acids could be decarboxylated to react with electrophiles under mild conditions. Second, Pd-mediated decarboxylation of vinyl carboxylic acids is also a feasible process because its activation barrier is only +26.5 kcal/mol. Third, alkanecarboxylic acids are probably not going to be useful in Pdcatalyzed decarboxylative couplings because of the very high

Factors Controlling Decarboxylative Coupling

ARTICLES

Table 3. Carboxylic Acid Substrate Effect on the Decarboxylation Step (kcal/mol)

a ∆Gqdecarboxylation ) ∆G(CP0 f CP1) + ∆G(CP1 f CP2) + ∆Gq(CP2 f TS1). compound that is generated after decarboxylation of the carboxylic acid, see also ref 34.

activation barriers (over +33 kcal/mol) and the propensity of alkyl-Pd to undergo side reactions. The above results appear to indicate that the activation barrier of Pd-mediated decarboxylation of RCOOH should correlate with the acidity of the R-H bond. Specifically, a more acidic R-H bond should be associated with a more readily decomposable R-COOH. This speculation is largely correct as shown by the plot of the decarboxylation barriers against the gas phase acidities of the Cipso-H bonds (Figure 6). The finding is important because it implies that transition-metal-mediated decarboxylation and C-H bond activation may share some common features. Despite the above correlation, it is surprising to notice that benzoic acids with electron-withdrawing substituents dramatically deviate from the correlation line (Figure 6). In fact, an electrondonating group (e.g., OMe) reduces the activation barrier of decarboxylation whereas an electron-withdrawing substitutent (e.g., NO2) does the opposite. This theoretical prediction is in agreement with the experiment, where electron-rich benzoic acids were also found to be better substrates in the decarboxylative couplings.9,10 (34) C-H acidities in Table 3 are theoretical values at the B3LYP/6311+G(d,p) level. These theoretical values are found to reproduce the available experimental data fairly well: (a) Romer, B.; Gatev, G. G.; Zhong, M.; Brauman, J. I. J. Am. Chem. Soc. 1998, 120, 2919. (b) Fattahi, A.; McCarthy, R. E.; Ahmad, M. R.; Kass, S. R. J. Am. Chem. Soc. 2003, 125, 11746, and references cited therein.

b

Gas phase acidity of the Cipso-H bond of the hydrocarbon

Figure 6. Correlation between the activation barrier of Pd-mediated decarboxylation of RCOOH and the acidity of the R-H bond.

Detailed analysis indicates that the problem stems from the DMSO removal step (i.e., CP1 f CP2). For instance, for 4-NO2C6H4COOH the removal of a DMSO from CP1 to generate CP2 costs +14.7 kcal/mol, whereas the same process only costs +6.7 kcal/mol for 4-OMe-C6H4COOH. The large difference between the two acids can be explained by the formation of the Cipso-Pd(II) bond in CP2. Specifically, a 4-NO2 group (as a π acceptor) should J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

645

ARTICLES

Zhang et al.

Table 4. Metal Center Effect on the Decarboxylation Step (kcal/mol)a

a

metal

∆G(CP0 f CP1)

∆G(CP1 f CP2)

∆Gq(CP2 f TS1)

∆Gqdecarboxylation

Ni Pd Pt

10.0 3.7 9.7

13.6 7.8 8.4

19.9 17.5 16.5

+43.5 +29.0 +34.6

∆Gqdecarboxylation ) ∆G(CP0 f CP1) + ∆G(CP1 f CP2) + ∆Gq(CP2 f TS1).

destabilize the Cipso-Pd(II) bond that is perpendicular to the phenyl ring (see Figure 1), but this π(NO2)-σ(C-Pd) type of interaction does not play any role in determining the R-H acidity (because the R-H bond is not perpendicular to the phenyl ring). As a result, substituted benzoic acids do not rigorously follow the correlation in Figure 6. Furthermore, it is interesting to observe that the activation barrier of the 4-OMe-substituted benzoic acid (+25.6 kcal/mol) is about 3 kcal/mol higher than that for 2-OMe-substituted benzoic acid (+22.4 kcal/mol). Similarly, the activation barrier of the 4-NO2substituted benzoic acid (+30.7 kcal/mol) is about 5 kcal/mol higher than that for 2-NO2-substituted benzoic acid (+25.2 kcal/ mol). These observations indicate that the ortho substituents influence the reactivity of benzoic acids more via their influence on the σ electron density than the π electron density. Similar effects were reported by Goossen and Thiel in their study on Cu-catalyzed decarboxylation.18 This effect also explains why 2-furancarboxylic acid has a very low activation barrier for decarboxylation as compared to benzoic acid. 4.5. Metal Center Effect on Decarboxylation. All the d10 group elements (Ni, Pd, and Pt) are examined for the decarboxylation process. According to the results in Table 4, the overall activation barrier for the extrusion of CO2 from benzoic acid follows the following order: Pd < Pt < Ni. As to Ni catalyst, it is found that the energy cost for the CO2 extrusion step (CP2 f TS1, +19.9 kcal/mol) is only 2.4 kcal/mol higher than the Pd case. However, carboxyl exchange (CP0 f CP1) and removal of one DMSO (CP1 f CP2) cost a considerable amount of free energy for Ni. As to Pt, the energy cost for the decarboxylation step (CP2 f TS1, +16.5 kcal/mol) is even 1.0 kcal/mol lower than the Pd case. Unfortunately, the energy cost for the carboxyl exchange step (CP0 f CP1) is fairly high for the Pt catalyst. One may propose that the unfavorable effects of carboxyl exchange and DMSO removal steps in Ni- and Pt-catalyzed decarboxylation could be overcome by changing the ligands. It remains interesting to see whether Niand Pt-catalyzed decarboxylative cross-coupling could be accomplished experimentally. 5. Conclusions

Transition-metal-catalyzed decarboxylative cross-coupling uses readily available carboxylic acids to replace unstable organometallic reagents. This method presents a new and important direction in

646

J. AM. CHEM. SOC.

9

VOL. 132, NO. 2, 2010

synthetic organic chemistry. In the present work, we report the first comprehensive theoretical study on the mechanism of Pdcatalyzed decarboxylative Heck reaction discovered by Myers and co-workers. It is found that the overall catalytic cycle is composed of four steps: decarboxylation, olefin insertion, β-hydride elimination, and catalyst regeneration. Decarboxylation is concluded to be the rate-limiting step, in which Pd(II) mediates the extrusion of CO2 from an aromatic carboxylic acid to form a Pd(II)-aryl intermediate. The dissociative pathway is favored in decarboxylation, so that dissociation of one DMSO ligand must take place prior to the extrusion of CO2. Further analysis on the various factors that may control the efficiency of Pd-catalyzed decarboxylative Heck reactions yields the following conclusions. First, PdCl2 and PdBr2 are much worse catalysts than Pd(TFA)2, because the exchange of Cl or Br in a Pd complex with the carboxylate substrate is a highly unfavorable step in thermodynamics. Second, DMSO presents the best compromise between the carboxyl exchange and decarboxylation steps. Phosphine and N-heterocarbene ligands disfavor the decarboxylation step because of their electron richness. Pyridine-type ligands, on the other hand, disfavor the carboxyl exchange step. Third, there is a good correlation between the activation barrier of R-COOH decarboxylation and the acidity of the R-H bond. A more acidic R-H bond causes a more readily decomposable R-COOH. However, substituted benzoic acids show deviation from the correlation because of the involvement of π(substituent)σ(Cipso-Pd) interaction. Finally, Ni and Pt are worse catalysts than Pd for decarboxylation because of the less favorable carboxyl exchange and/or DMSO removal steps in Ni and Pt catalysis. These results should also have valuable implications for other transitionmetal-catalyzed decarboxylative reactions. Acknowledgment. This study was supported by the National Natural Science Foundation of China (Nos. 20832004, 20802040, 20972148), the Specialized Research Fund for the Doctoral Program of Higher Education (No. 200800030074), and the NCET (080519). Supporting Information Available: Detailed optimized geometries, free energies, thermal corrections. Discussion on the binding mode of DMSO. Full citation of ref 19. This material is available free of charge via the Internet at http://pubs.acs.org. JA907448T